Online Program Home
My Program

Abstract Details

Activity Number: 259 - SPEED: Missing Data and Causal Inference Methods, Part 2
Type: Contributed
Date/Time: Monday, July 29, 2019 : 3:05 PM to 3:50 PM
Sponsor: Health Policy Statistics Section
Abstract #307639
Title: Developing and Evaluating Methods to Impute Race/Ethnicity in an Incomplete Dataset
Author(s): Gabriella Silva* and Amal N. Trivedi and Roee Gutman
Companies: Brown University and Brown University and Brown University
Keywords: Race/ethnicity; Health care; Bayesian regression; Multiple imputation; Fully conditional specification
Abstract:

Patient self-reported racial/ethnic information is needed to identify and address racial/ethnic disparities in the health care system; however, this information is often missing. Recent studies have focused on developing indirect methods for estimating race/ethnicity but these usually only consider geocoded and surname data as predictors, tend to perform poorly among racial minorities, and fail to provide race estimates for subjects missing some of this information. Our objective was to address these limitations by developing a novel method for imputing race/ethnicity using data from Rhode Island Medicaid beneficiaries. Current race estimation methods and newly developed ones were compared under various missing data mechanisms using racial composition estimates and area under the ROC curve statistics. We found that family race is an important predictor and that Bayesian regression models (BRM) provide better estimates than previously proposed methods. Missing race was multiply imputed using joint modeling (JM) and fully conditional specification (FCS). Across all missing data mechanisms, post-imputation analyses showed that FCS with a BRM is superior to JM for race imputation.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2019 program